Remaining useful life estimation based on wavelet decomposition: Application to bearings in reusable liquid propellant rocket engines
Résumé
The increasing trend towards the development of reusable rocket engines pushes towards the development of Prognosis and Health Monitoring (PHM) approaches useful to monitor the health of the system and plan its maintenance. In this perspective, we present a hybrid method for Remaining Useful Life (RUL) estimation of rocket engines turbo-pump bearings based on Maximum Overlap Discrete Wavelet Packet Transform (MODWPT) and polynomial approximation. The proposed method calculates the bearing RUL in six main steps: data acquisition, wavelet decomposition, feature extraction, degradation detection, Health Indicator (HI) computation and RUL estimation. The obtained results showed that this technique can be successfully used to separate and isolate vibration trends connected to progressive degradation. The main contribution of this work consists in proposing a monotonically increasing HI which maintains a physical meaning allowing to estimate the degradation level. The proposed HI has proven to be able to effectively quantify the level of degradation and predict it for a certain group of degradation evolution profiles.